Growth | Scaling Your Startup S2 E1 with Craig Zingerline and Allen Chen | E1198

[@DarshanKalola]

I love how Jason gets both of them to be specific-like 2-3x, 70%-80% instead of just more or less. I know specifics can't capture complexity but they are very valuable to listeners.

[@bigfernhead]

Love the questions!

[@screwfaceclub]

5:08 in and loving it.

[@screwfaceclub]

His voice is a bit like Startup ASMR

[@richwu]

At what point are funnel metrics meaningful? For example, sample size my be very small in the early stages and metrics like conversion and core retention may fluctuate wildly week-to-week. As a corollary, how do you ensure your A/B tests are statistically significant when your sample size is relatively small?

[@craigzingerline8332]

Richard - great questions. Core retention will be the last metric you probably use as an indicator of success since it takes longer to normalize. But your proxy metrics can be used to determine if the core actions that you believe will support retention are being used by your users, and ideally getting better over time. Early on though you are right - metrics are all over the place. With A/B testing early on I look for 2 things: 1) if sample size is small, how much risk in terms of false positives or negatives are you willing to take? I'll go for a P-value or confidence level at or around 80%, which leaves a 20% margin of error. 2) directionally does what your doing get you positive qualitative feedback from users - if so, you're possibly onto something. Take it all with a grain of salt early though.

[@richwu]

@@craigzingerline8332 Thanks for the insight: it seems there is a factor of risk with any experiment (e.g., 20% false positives), but in the long run metrics should improve up and to the right.

[@craigzingerline8332]

@@richwu that's the goal!

[@xxthesayborxx]

Sería bueno que si entrevistas al ceo de Latinoamérica tendrás más suscriptores